# Anthropic Looks to Buy Decart for $6 Billion

An episode of [Bloomberg Tech](https://invisiblenetwork.ai/podcast/1611767434), 13 Aug 2026, 47 min.

Audio: https://podtrac.com/pts/redirect.mp3/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/41764a4f-fc64-4e11-89ba-ae7c0030ab5e/96e30deb-1485-4662-8b7d-b4a50118c3cc/audio.mp3?utm_source=Podcast&in_playlist=9caafc41-289c-4115-995d-ae7c0030ab75

Bloomberg’s Ed Ludlow breaks down what could be Anthropic's biggest acquisitions yet: a $6 billion deal for AI startup Decart to help it squeeze more performance out of its computing infrastructure. Plus, CoreWeave warns investors that shifting from its exclusive use of Nvidia AI chips may cost time and money, and Cerebras' second-quarter earnings send its shares sliding after underwhelming growth. See omnystudio.com/listener for privacy information.

## Transcript

[0:02] Speaker 1: Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is live
[0:14] Speaker 1: from the heart of Silicon Valley with Ed La Though
[0:18] Speaker 1: in Fan Francisco.
[0:22] Speaker 2: This is Bloomberg Tech coming up.
[0:23] Speaker 3: Anthropic is reportedly eyeing one of its biggest acquisitions yet,
[0:27] Speaker 3: a six billion dollar deal for AI's startup dcart, to
[0:30] Speaker 3: help it squeeze more performance out of its computing infrastructure
[0:33] Speaker 3: plus core. We've One's investors that shifting from its exclusive
[0:37] Speaker 3: use of Nvidia AI chips is going to cost them
[0:40] Speaker 3: time and money, and Cerebras also out with results ches
[0:43] Speaker 3: sliding after underwhelming growth. We're going to talk to the
[0:46] Speaker 3: CEO later this hour. Let's get right to our top story.
[0:50] Speaker 3: Bloomberg's learned Anthropic is in talks to buy AI world
[0:53] Speaker 3: model startup dcart Ai for about six billion dollars. According
[0:57] Speaker 3: to sources, it would be Nthropic's biggest acquisition. Yeah, Bloomberg
[1:00] Speaker 3: AI reporter Shriigafari Bloomberg Deals reporter Ryan Gould both with
[1:04] Speaker 3: US now and actually Shrein, I want to start with you. Yes,
[1:07] Speaker 3: there's the world model's element of this, but Decot does
[1:10] Speaker 3: something else that will be appealing to Anthropic as it
[1:13] Speaker 3: hopes to get the most out of its compute.
[1:16] Speaker 4: Right, and that's basically maximizing the efficiency out of chips,
[1:20] Speaker 4: and now just in video chips, but other types of
[1:22] Speaker 4: chips like TPUs and Amazon's tranium chips, and so that
[1:27] Speaker 4: is important in a time when obviously Anthropic and the
[1:32] Speaker 4: AI vendors are in a crunch trying to be able
[1:35] Speaker 4: to keep up with demand and offer all the compute
[1:38] Speaker 4: they can get.
[1:39] Speaker 3: Ryan take us into the deal here we're saying that
[1:42] Speaker 3: this would be Anthropic's biggest acquisition. To me, six billion
[1:45] Speaker 3: dollars seems quite modest in the landscape of deals we've
[1:48] Speaker 3: been covering of late, But of course around the corner
[1:52] Speaker 3: is an expected.
[1:52] Speaker 5: IPO exactly ed and I think you know it is
[1:55] Speaker 5: or would be an topic's largest known deal today, and
[1:58] Speaker 5: who knows what they've been doing behind the means. I
[2:01] Speaker 5: think it's just super interesting in terms of timing, as
[2:03] Speaker 5: you say that IPO coming is potentially as soon as October,
[2:07] Speaker 5: which is, you know, really not that far away at all.
[2:09] Speaker 5: I think when you put this into perspective, Oranthropic doing
[2:13] Speaker 5: this kind of deal as a private company is perhaps
[2:15] Speaker 5: a little bit easier than as a public one, and
[2:17] Speaker 5: I think it might actually help them to communicate that
[2:20] Speaker 5: story at least on the cost efficiency side a little better.
[2:23] Speaker 5: When it comes to you know, that finalized S one.
[2:26] Speaker 5: They did obviously file confidentially, but this acquisition itself would
[2:29] Speaker 5: not obviously show up in that existing filing. I think
[2:33] Speaker 5: six billion is modest, but you know, you've seen this
[2:36] Speaker 5: company was just it just raised a four billion dollar
[2:38] Speaker 5: valuation not so long ago that that round was backed
[2:41] Speaker 5: by you know, in Vidia, And I think if we
[2:44] Speaker 5: just pause a little bit on that. I think I
[2:45] Speaker 5: saw someone make a point online this morning that you know,
[2:48] Speaker 5: Anthropic itself is starting to look a little more like
[2:50] Speaker 5: in Nvidia as well as being an Nvidia competitor. And
[2:53] Speaker 5: I think to put that into like the sum of things,
[2:56] Speaker 5: who knows how that'll be valued to you know, how
[2:59] Speaker 5: investors will credit that At the time of IPO.
[3:02] Speaker 3: Serene, you cover this company as closely and as deeply
[3:05] Speaker 3: as any journalists on the planet. I really love to
[3:07] Speaker 3: understand what their M and A story has been to day.
[3:11] Speaker 3: You know, they are a frontier lab born out of
[3:13] Speaker 3: research working on frontier models. Have they gone out shopping
[3:18] Speaker 3: to buy the other bits that have completed their business.
[3:22] Speaker 4: Anthropic has been quite disciplined both in their early kind
[3:26] Speaker 4: of compute efforts. You had Anthropic CEO Dario m and
[3:29] Speaker 4: A saying he didn't want to sort of yolo and
[3:32] Speaker 4: go too far with over extending on compute.
[3:35] Speaker 6: They've similarly been more disciplined than some of their.
[3:37] Speaker 4: Competitors on their m and A strategy so far, and
[3:40] Speaker 4: haven't made as many splashy big deals. If you think
[3:42] Speaker 4: to open Aye buying Johnny, I've startup, or other major headlines,
[3:47] Speaker 4: Anthropics have been a little more under the radar, so
[3:49] Speaker 4: this wouldmark, as we reported, their biggest acquisition yet and
[3:53] Speaker 4: may represent a shift where as Ryan was setting saying
[3:56] Speaker 4: ahead of a potential blockbuster IPO, they are starting to
[4:00] Speaker 4: make some of these bigger transactions in areas where they.
[4:02] Speaker 6: May need help.
[4:04] Speaker 3: Ryan Anthropic has raised I think more than one hundred
[4:08] Speaker 3: billion dollars just a year to date in those two
[4:10] Speaker 3: mega rounds that they did. I'm sure most of that's
[4:12] Speaker 3: on the balance sheet. Dcart, It's reported was looked at
[4:15] Speaker 3: by Musk and space XAI. What are you seeing across
[4:19] Speaker 3: the field at the moment. This is an active space.
[4:22] Speaker 3: There are lots of transactions happening at all kinds of levels.
[4:25] Speaker 2: I think that's true.
[4:26] Speaker 5: I think you saw that report out of Israel earlier
[4:29] Speaker 5: on this week that I think put SpaceX's name next
[4:31] Speaker 5: to dcart Elon obviously came out and denied that report.
[4:35] Speaker 5: I think what I would say taking a step back
[4:37] Speaker 5: about Antropic in particular is we know and we have
[4:39] Speaker 5: spoken to many people are on the sidelines over the
[4:42] Speaker 5: past few months that they have been quite actively shopping
[4:44] Speaker 5: for acquisitions, obviously putting in this in the context as
[4:48] Speaker 5: well as them trying to build out their own chip effort.
[4:50] Speaker 5: They were looking for something on the hardware layer two
[4:52] Speaker 5: earlier this year. I think this is a company that's
[4:54] Speaker 5: really trying to make that vision a lot clearer again,
[4:58] Speaker 5: as I say, ahead of the IPO, and when you
[5:00] Speaker 5: look at across the street and think about some of
[5:02] Speaker 5: these other sort of chip influence type players, efficiency players
[5:05] Speaker 5: that are coming out on the market, like you know
[5:08] Speaker 5: we've been covering, tends to try and earn you know
[5:09] Speaker 5: it well.
[5:10] Speaker 2: These are all types of deals where.
[5:12] Speaker 5: I think companies like Anthropic are realizing that there is
[5:15] Speaker 5: actually massive value here and there are still quite few
[5:18] Speaker 5: of them at scales, so it's kind of a scarcity player.
[5:21] Speaker 3: I think at this point in each of those cases,
[5:23] Speaker 3: they've gone after set ups to have a very clear specialism,
[5:26] Speaker 3: very clear niche. The teams updated the story on the
[5:29] Speaker 3: Bloomberg to go and read it in full Bloomberg, Shringafari
[5:31] Speaker 3: and Ryan Gould, thank you very much. Indeed, let's go
[5:34] Speaker 3: to public markets. The S and P five hundred is
[5:37] Speaker 3: just at fresh record highs and like a lot of
[5:39] Speaker 3: that is the story here today, Storage Memory, some of
[5:43] Speaker 3: the high flying chip names.
[5:44] Speaker 2: But our guess, the next guest is.
[5:46] Speaker 3: You don't have to own the megacaps to play the
[5:49] Speaker 3: AI boom. She sees opportunities in mid caps from the
[5:52] Speaker 3: companies powering the build out to those actually using AI
[5:56] Speaker 3: to boost productivity. Lorrie Keith, Finasis Investments, port Pholio managers
[5:59] Speaker 3: see the senior analyst joins us. Now there's picks and shovels,
[6:03] Speaker 3: but picks and shovels can be big or small or
[6:06] Speaker 3: somewhere in between. And we've become so focused on this
[6:09] Speaker 3: small group of ginormous technology companies we talk less about
[6:13] Speaker 3: the rest of the field.
[6:14] Speaker 2: And that's the field that you play in.
[6:16] Speaker 7: That's correct.
[6:17] Speaker 2: Thanks Sad.
[6:17] Speaker 8: It's great to be here and absolutely, I mean we're
[6:19] Speaker 8: seeing really this opportunity broadening out beyond just a small
[6:23] Speaker 8: group of players which have really powered the CAPEX trade
[6:26] Speaker 8: so far, and we're seeing that broaden out to many
[6:29] Speaker 8: of the companies that maybe aren't even thought of as
[6:31] Speaker 8: AI companies per se. They're not known as technology companies.
[6:35] Speaker 8: But you're seeing the value chain expand beyond just the
[6:39] Speaker 8: compute the memory players into areas like water and power.
[6:43] Speaker 8: I mean, these are critical areas when you look beyond
[6:45] Speaker 8: the value chain that are going to be needed to
[6:47] Speaker 8: support the growth of AI and data centers. And so
[6:51] Speaker 8: there's a number of really interesting opportunities I think that
[6:53] Speaker 8: sit beneath the surface in the mid cap arena that
[6:57] Speaker 8: we find very attractive.
[6:58] Speaker 3: Really interesting how that's played out over time. If you
[7:02] Speaker 3: take high bandwidth memory as the example, it's been quite
[7:06] Speaker 3: some time where high bandwidth memory is part of the
[7:08] Speaker 3: server design, and it seemed like overnight we were like, oh,
[7:11] Speaker 3: my goodness, everyone needs hie bandwidth memory. Is that analogous
[7:15] Speaker 3: with what's happening in power and water? People are saying, oh, oh,
[7:17] Speaker 3: we better do something.
[7:18] Speaker 7: I think it is.
[7:19] Speaker 8: I mean, there's obviously a lot more challenges as you
[7:22] Speaker 8: move from the law of physics around implementing some of
[7:25] Speaker 8: the power needs that are required to support all these
[7:27] Speaker 8: data centers, and so there's I think a whole breadth
[7:30] Speaker 8: of opportunities that aren't necessarily as focused on. But when
[7:33] Speaker 8: you think about not only the data centers around the
[7:38] Speaker 8: need for power and energy, but also water. When you
[7:41] Speaker 8: look at the scarcity of water, the water treatment that's
[7:45] Speaker 8: required all the way from the actual development of the
[7:48] Speaker 8: chips into the data centers for cooling, that's another really
[7:52] Speaker 8: interesting area that I think as an enabler, those companies
[7:55] Speaker 8: that are providing those technologies are going to be very
[7:57] Speaker 8: well positioned when you look out or the next to
[8:00] Speaker 8: ten years.
[8:01] Speaker 3: When you look at the mag seven there's a lot
[8:03] Speaker 3: that's analogous between them. They are hyperscalers, or they have
[8:07] Speaker 3: similar business models, or they rely on the same partner
[8:10] Speaker 3: and video. In the mid cap segment, there's a lot
[8:13] Speaker 3: more diversity, very much, so it's a bigger feel like that,
[8:17] Speaker 3: how do you do how do you parse all of that?
[8:20] Speaker 8: Yeah, it's a great point, and I would say within
[8:23] Speaker 8: the large cap arena, when you look at the concentration
[8:26] Speaker 8: in the index, and the large cap index roughly about
[8:29] Speaker 8: forty percent of the index is driven by ten companies.
[8:32] Speaker 8: When you look at MidCap, the largest constituent is actually
[8:35] Speaker 8: closer at about seventy to eighty basis points, so a
[8:37] Speaker 8: lot more dispersion, a lot more opportunities that And I'll
[8:41] Speaker 8: also say you noticed, from a valuation perspective, you've got
[8:44] Speaker 8: large caps trading today close to twenty one times, you've
[8:46] Speaker 8: got small taps closer to twenty five times, and you've
[8:50] Speaker 8: got midcaps around seventeen times. And this is a very
[8:53] Speaker 8: attractive value prop relative to comparable earnings growth that we
[8:57] Speaker 8: see over the next few years. And so we think
[9:01] Speaker 8: there's a lot of great opportunities. Again, and also the
[9:04] Speaker 8: concentration the sector composition is quite different than large caps,
[9:08] Speaker 8: so you're able to diversify away from this very what
[9:13] Speaker 8: is becoming a more crowded trade into areas that will
[9:17] Speaker 8: benefit from AI, but they're not dependent upon AI. So
[9:20] Speaker 8: a lot of the opportunities that we focus on have
[9:22] Speaker 8: AIS and enhancement.
[9:23] Speaker 2: Whether you give a case study.
[9:25] Speaker 8: Absolutely, so let me start with let me start with
[9:28] Speaker 8: a company like Hubble, which is one of the leading
[9:31] Speaker 8: providers of equipment that goes into the grid and supporting
[9:36] Speaker 8: the build out. That opportunity set was already there before AI,
[9:40] Speaker 8: as there's a need to modernize the grid as you're
[9:42] Speaker 8: diversifying your power sources transitioning away from coal, So that
[9:46] Speaker 8: is actually getting enhanced by the need for more data
[9:49] Speaker 8: centers and the need for more power. So that's a
[9:52] Speaker 8: great opportunity on the user side, though I would also
[9:55] Speaker 8: point to even a company.
[9:57] Speaker 2: Like a robin Hood.
[9:58] Speaker 8: Robin Hood, actually there's all whole host of companies when
[10:01] Speaker 8: you look out where's the next spectrum of opportunity. Those
[10:04] Speaker 8: are companies that are able to leverage AI to improve
[10:08] Speaker 8: their business outcomes, to improve their competitive vantages. In terms
[10:13] Speaker 8: of Robinhood, they are now powering about seventy five percent
[10:17] Speaker 8: of their customer service calls using AI, so that is
[10:21] Speaker 8: leading to very tangible benefit as well as incorporating that
[10:24] Speaker 8: into engineering those Certainly it's early days on the user
[10:28] Speaker 8: side from an enterprise perspective, but we think there's a
[10:30] Speaker 8: really long tail as those users start to adopt AI
[10:35] Speaker 8: and strengthen those competitive vantages, we should see that over
[10:38] Speaker 8: time broad and beyond the use cases around customer service
[10:42] Speaker 8: and engineering that most companies have focused on so far
[10:45] Speaker 8: into things that helped to really drive margin.
[10:48] Speaker 3: We saw this conversation saying the S and P five
[10:50] Speaker 3: hundreds at another fresh record high.
[10:52] Speaker 2: I mean starts go up and down. As of course
[10:54] Speaker 2: that one.
[10:54] Speaker 3: Of the biggest contributors or the best performers of States
[10:57] Speaker 3: Western Digital. Yes, and that's probably the is a more
[11:00] Speaker 3: crowded trade. You know, the story is quite well known
[11:03] Speaker 3: at this point, but it is a name that that
[11:05] Speaker 3: you hold and that you think about where does it
[11:08] Speaker 3: fit in your broader thesis.
[11:09] Speaker 8: I think absolutely it's a key enabler as you look
[11:12] Speaker 8: at the need for storage as.
[11:14] Speaker 6: AI, the amount of data, the amount.
[11:16] Speaker 8: Of compute that's driving the lllms that are driving new data.
[11:21] Speaker 8: The hard just drive business will continue to be really
[11:25] Speaker 8: relevant and we think there's a long secular tale for that.
[11:27] Speaker 8: So we've actually held that for a while for the
[11:30] Speaker 8: last couple of years, but we still still see a
[11:32] Speaker 8: nice runway for growth given the opportunity there.
[11:37] Speaker 3: Lori Pief for Permasa Sevestment has been great to have
[11:39] Speaker 3: you here in the studio and Essa, thank you very
[11:41] Speaker 3: much so much. Indeed, another start that we're watching today
[11:44] Speaker 3: is cerebrous and the stop is under pressure.
[11:47] Speaker 2: You know.
[11:48] Speaker 3: It's another high expectations earning story.
[11:52] Speaker 2: The stockdown fourteen percent.
[11:54] Speaker 3: There's questions about the lumpiness of hardware sales and why
[12:00] Speaker 3: renting out compute capacities slightly eclipsing that. We're gonna speak
[12:03] Speaker 3: to the CEO, Andrew Feldman later in the out.
[12:08] Speaker 2: Okay, coming out. We also have more earning.
[12:10] Speaker 3: Cisco reporting fourth quarter results disappointing investors despite beating estimates
[12:15] Speaker 3: in the quarter gone. It's all about the AI revenue outlook,
[12:17] Speaker 3: which we'll go through next.
[12:18] Speaker 2: This is Bloomberg Tech.
[12:32] Speaker 3: Cisco reported fourth quarter earnings, disappointing investors after the company
[12:36] Speaker 3: projected it would make seven point five billion dollars for
[12:39] Speaker 3: sales tied to the AI data center Boom, a number
[12:42] Speaker 3: investors deemed conservative compared to prior figures like nine point
[12:47] Speaker 3: three billion dollars over the past year. Bloomberg Intelligence is
[12:49] Speaker 3: wu jinho joins us.
[12:51] Speaker 2: Now.
[12:51] Speaker 3: The problem is reconciling a backlog with forecasted AI revenue
[12:56] Speaker 3: and how much of overall revenue is AI revenue?
[13:00] Speaker 2: Where did your math? Now?
[13:02] Speaker 9: Yeah, so thanks for having me on look. I think
[13:06] Speaker 9: the AI revenue of seven point five billion was quite fine.
[13:09] Speaker 9: You know, they already guided to six billion in the
[13:12] Speaker 9: last quarter results of fiscal twenty twenty seven. You know,
[13:14] Speaker 9: I think the hang up with investors, quite frankly ed,
[13:17] Speaker 9: is more on the gross margin side than anything else.
[13:20] Speaker 9: They did sixty six percent gross margins in the fourth quarter,
[13:23] Speaker 9: guide it to sixty six percent in the first quarter.
[13:25] Speaker 9: But it seems like there's going to be like one
[13:28] Speaker 9: hundred and fifty point basis drop off in gross margins
[13:30] Speaker 9: in quarters two to three. So the earnings power isn't
[13:34] Speaker 9: as strong as people at hope for the full year.
[13:36] Speaker 10: Right.
[13:38] Speaker 3: The nine point three billion dollar figure I reference was
[13:40] Speaker 3: the orders for AI related gear. That it's a massed
[13:44] Speaker 3: over the last year, and so they're saying, Okay, you've
[13:46] Speaker 3: told us here's a backlog, and here's how much revenue.
[13:49] Speaker 3: Is it a conversion rate, it's to you, Yeah, there
[13:52] Speaker 3: is some conversion rate.
[13:54] Speaker 9: Look, the timing of the conversion rate is a little
[13:57] Speaker 9: bit unclear as well, which which may hold back some
[14:02] Speaker 9: of that and some of those sales is going into
[14:06] Speaker 9: fiscal twenty twenty eight as well, right, And the issue
[14:10] Speaker 9: is is that it's not going to be It's not
[14:12] Speaker 9: unique to Cisco. We're hearing the same thing on all
[14:14] Speaker 9: the optical players. We're also hearing the same thing on
[14:17] Speaker 9: the semicon players as well, primarily because of the supply tightness.
[14:22] Speaker 3: You know, Chuck Robbins, right, I think he said this
[14:25] Speaker 3: was prudence as opposed to conservative. Does it matter that
[14:29] Speaker 3: the market interprets a set of numbers as conservative?
[14:34] Speaker 9: Yeah, Look, I think given the big run up, like
[14:38] Speaker 9: the stock has been up over fifty percent a year
[14:40] Speaker 9: to date, expectations have been heading into the print, and
[14:45] Speaker 9: I think they were expecting higher numbers as on the
[14:48] Speaker 9: gross margin side, earnings as well as on sales. When
[14:52] Speaker 9: they report a twenty seven guide.
[14:55] Speaker 3: Why are expectations so high that Cisco is going to
[14:58] Speaker 3: be a winner along with everyone else in this field.
[15:01] Speaker 7: Yeah.
[15:02] Speaker 9: If you think about like the two hottest spaces in
[15:07] Speaker 9: the AI play right aside from GPUs optics as well
[15:11] Speaker 9: as some of the networking silicon, Cisco has both of them,
[15:16] Speaker 9: and they're very well positioned in both of them. And
[15:19] Speaker 9: the optics play, I mean, we're not only seeing one
[15:22] Speaker 9: hundred percent growth, We're seeing about one hundred and fifty
[15:25] Speaker 9: percent growth, right, and Cisco has been a number one
[15:28] Speaker 9: market share leader there and on the silicon side, they're
[15:31] Speaker 9: deeper penetration there. So expectations of Cisco doing well on
[15:35] Speaker 9: both ends and the hyperscale cloud side have been fairly
[15:38] Speaker 9: high heading into the print, so you know, and they
[15:41] Speaker 9: expect that to move on going forward.
[15:44] Speaker 3: Blueberg Intelligence is wou Jinhe great to have you back
[15:47] Speaker 3: on Bloomberg Tech.
[15:47] Speaker 2: Thank you very much.
[15:49] Speaker 11: Call.
[15:49] Speaker 3: We've warned its investors that moving away from its exclusive
[15:53] Speaker 3: use of Nvidio chips will cost the company quote, time,
[15:56] Speaker 3: investment and resources.
[15:58] Speaker 2: That was in a regulatory filing.
[15:59] Speaker 3: A chain driven in anticipation by strong customer demand, belie
[16:04] Speaker 3: the most.
[16:04] Speaker 2: Brady Ford joins us.
[16:05] Speaker 12: Now.
[16:06] Speaker 3: Corey's a neocloud and one percent of its existing customers
[16:09] Speaker 3: rely on compute that is based on Nvidia architecture. The
[16:12] Speaker 3: problem is all those customers are thinking about life after
[16:16] Speaker 3: in Vidio, right.
[16:17] Speaker 2: That's the story here exactly.
[16:20] Speaker 13: Most people see neo clouds like Corey these kind of
[16:23] Speaker 13: vassal states of Nvidia. They sell their tech to some extent,
[16:27] Speaker 13: they do what they're told by Nvidia, and so any
[16:31] Speaker 13: sign of daylight between the large neo clouds and Nvidia
[16:35] Speaker 13: is quite interesting to us because it just suggests that
[16:39] Speaker 13: are the chips shifting currently in Vidia is the undisputed
[16:44] Speaker 13: king for AI chips. Of course, you ask any big customer, Meta, Microsoft, Amazon,
[16:49] Speaker 13: any of them. They hope that's not the world they're
[16:52] Speaker 13: in in a couple of years, and so it will
[16:54] Speaker 13: be interesting to see if that plays out. How do
[16:56] Speaker 13: the neo clouds adapt.
[16:58] Speaker 3: Right in Video also owns ten percent of core Weave,
[17:02] Speaker 3: effectively his core We've ever said anything about the future
[17:06] Speaker 3: silicon or compute platform plan, like are they in talk
[17:09] Speaker 3: so with anyone else other than Nvideo or is this
[17:11] Speaker 3: them just saying like, heads up, this is a possible risk.
[17:16] Speaker 13: They've been asked a few times directly, most recently I
[17:19] Speaker 13: could find in December, and they pretty much said what
[17:22] Speaker 13: people want is in video and we're going to keep
[17:24] Speaker 13: giving it to them. But this is a change in
[17:27] Speaker 13: their language, right, And I think monitoring these small changes
[17:30] Speaker 13: in language, clearly somebody at core Weave was saying, hey,
[17:34] Speaker 13: let's change this for a reason, and I think it's
[17:37] Speaker 13: a very interesting clue about how the chips may be
[17:39] Speaker 13: shifting in the coming quarters real quick.
[17:42] Speaker 2: Brody earnings earlier this week. We're good for core weave.
[17:46] Speaker 6: Oh.
[17:47] Speaker 13: I mean, if you're a coreve investor, you had a
[17:48] Speaker 13: beautiful week. You know, rev was up, the margin story
[17:52] Speaker 13: was beautiful, and so yeah, the bearishness against neo clouds
[17:56] Speaker 13: has dissipated a bit this week.
[17:59] Speaker 3: All right, Blimbeks, Brady Forward and all things no cloud,
[18:02] Speaker 3: thank you very much. Indeed, take a look at shares
[18:05] Speaker 3: of stub Hub lower today. Earnings forecast underwhelming. Pretty significant story,
[18:12] Speaker 3: but a similar story to what we've seen throughout this
[18:14] Speaker 3: earning period where the company went in with high expectations.
[18:18] Speaker 3: The stockdown thirteen percent, almost fourteen percent. That had been
[18:22] Speaker 3: down even more than that in the pre market session.
[18:24] Speaker 2: Coming up on the.
[18:25] Speaker 3: Show, SpaceX clears its first major lock up test. Why
[18:29] Speaker 3: the feared wave of selling never materialize? That's next, This
[18:33] Speaker 3: is Blimberg Tech. Let's get over to New York where
[18:48] Speaker 3: Bloomberg's your hirah and and is standing by Hiya hira
[18:50] Speaker 3: hi ed.
[18:51] Speaker 14: It's time out for tapping tech First up. Cxm T
[18:55] Speaker 14: is now China's most valuable company after surpassing ten cents
[18:59] Speaker 14: showing at the sky high demand for memory chipstock show
[19:03] Speaker 14: no signs of slowing down. The switch reflects investors continued
[19:07] Speaker 14: preference for AI linked hardware over traditional Internet champions. Plus Also,
[19:13] Speaker 14: China's largest ride hailing platform dd slung Bank back to
[19:17] Speaker 14: a profit after previously reporting back to back quarterly losses.
[19:21] Speaker 14: The company reported a net income of one one hundred
[19:24] Speaker 14: twenty eight million dollars, with revenue increasing eleven percent thanks
[19:28] Speaker 14: to China's strong right hailing demand, and SpaceX saw its
[19:33] Speaker 14: highest intra day level jump in over a month after
[19:36] Speaker 14: releasing a new AI model called rock Bot. The new
[19:40] Speaker 14: software is designed to handle a wide range of professional work,
[19:43] Speaker 14: including signing into various apps and websites and allowing bots
[19:47] Speaker 14: to share details and contexts with one another.
[19:50] Speaker 3: Ed all so good at coding and lombrang guitask, thank
[19:53] Speaker 3: you very much, you hire.
[19:54] Speaker 2: Let's stay on.
[19:54] Speaker 3: SpaceX shares had served thirty five percent post lock up
[19:58] Speaker 3: inspiring for yesterday's close, adding roughly five hundred billion dollars
[20:02] Speaker 3: in market value and defying fears that wave of insider
[20:05] Speaker 3: selling would pressure the stock. But several more lock up
[20:08] Speaker 3: expirations are ahead, investors are bracing for more volatility. Bloomberg's
[20:13] Speaker 3: calm Rhinicky joins us with what we need to know,
[20:15] Speaker 3: and you know, it's a calendar thing. In August there's more,
[20:19] Speaker 3: in October there's more, and next year there's more.
[20:22] Speaker 7: Yep.
[20:22] Speaker 15: It's something that we're going to just keep seeing, potentially
[20:25] Speaker 15: adding pressure to the stock, although going through the first
[20:28] Speaker 15: one it seems like investors didn't have a ton to
[20:30] Speaker 15: worry about. We saw pressure heading into the lock up,
[20:33] Speaker 15: especially with wednesdays earnings reports sort of coinciding right ahead
[20:37] Speaker 15: of when those shares did lock up, but that ended
[20:40] Speaker 15: up being the low for the stock for now. The
[20:43] Speaker 15: next day shares jumped and they rallied thirty five percent
[20:46] Speaker 15: over the next five days. We're seeing a little bit
[20:48] Speaker 15: of selling pressure today, but that's pretty expected after a
[20:51] Speaker 15: ten percent jump that we saw yesterday on that grock
[20:53] Speaker 15: bat news, So a lot of strength here. I think
[20:56] Speaker 15: it gave investors some confidence that there wasn't a lot
[20:59] Speaker 15: of selling with that first lockup. It was the biggest
[21:01] Speaker 15: one that we'll see, and it was more shares sort
[21:04] Speaker 15: of coming online than even actually in the IPO. So
[21:08] Speaker 15: a very good sort of first hurdle, as investors have said,
[21:11] Speaker 15: for the stock, but more pressure coming up in one
[21:14] Speaker 15: week we get the next lock up.
[21:16] Speaker 3: Expiration next summer is the big one, right, that's when
[21:19] Speaker 3: Elon Must shares subject to very specific lock up terms
[21:22] Speaker 3: hit the market.
[21:24] Speaker 15: Yes, you're right, So about a year after the IPO
[21:26] Speaker 15: is when that happens. And then sort of all of
[21:29] Speaker 15: the available shares will come online and we'll have the
[21:32] Speaker 15: total float, you know, for everyone to trade and we'll
[21:35] Speaker 15: see what happens. I mean, the first year of trading
[21:38] Speaker 15: post IPO is so volatile, and we know that there
[21:41] Speaker 15: are often large draw downs. I mean SpaceX has already
[21:44] Speaker 15: seen a huge draw down, you know, over a trillion
[21:47] Speaker 15: dollars in market value lost from you know, the top
[21:49] Speaker 15: a few days after the IPO through last week's bottom.
[21:52] Speaker 15: So lots of volatility here is to be expected continuing,
[21:56] Speaker 15: and I think we're going to.
[21:57] Speaker 3: See it in the year to come common really quickly.
[22:00] Speaker 3: Are two very large numbers Elon Must touted on the
[22:02] Speaker 3: Earning School. How investors thinking about that?
[22:05] Speaker 15: I think those are of paramount importance, right, we know
[22:07] Speaker 15: that so much of what investors are looking at here
[22:11] Speaker 15: is that future bid, Right, It's Elon's vision, and so
[22:14] Speaker 15: if you're a bull, that's what you're focused on. You're
[22:17] Speaker 15: saying the valuation doesn't matter. That's what I'm investing on.
[22:20] Speaker 15: If you're a bearer, you're a little bit more, you know,
[22:23] Speaker 15: scrutinizing that figure. So what those goalposts are very important.
[22:28] Speaker 3: One hundred billion RAR by year end, one trillion dollars
[22:31] Speaker 3: of revenue in twenty thirty. Bloomberg's coming, Ryanicky, thank you
[22:34] Speaker 3: very much, so coming up. IVP General partner kac Will
[22:37] Speaker 3: joins us. You're going to talk about investment opportunities in
[22:40] Speaker 3: the AI infrastructure layer, and probably a whole lot more.
[22:43] Speaker 3: It's halftime here in the beautiful city of San Francisco.
[22:48] Speaker 2: Stay with us. We'll be right back. This is Bloomberg Tech.
[22:59] Speaker 2: Welcome back to Bloomer Tech.
[23:00] Speaker 3: Stocks at record high, the S and P five hundred
[23:02] Speaker 3: at a fresh record high. Really, there's some economic data
[23:05] Speaker 3: underpinning that technology shares are pushing higher than as that
[23:09] Speaker 3: one hundred up per percentage point outperformance and chips. That's
[23:12] Speaker 3: kind of been the story through this middle earning cycle
[23:16] Speaker 3: where we've had a lot of the neo clouds. Of course,
[23:18] Speaker 3: we had the Biggin video announcement earlier in the week,
[23:20] Speaker 3: and that has carried the chip sector with some momentum
[23:24] Speaker 3: in private markets. There's a lot going on too. Let's
[23:26] Speaker 3: get today's big number. One hundred and ninety billion dollars.
[23:29] Speaker 3: That's Data Bricks's current valuation after raising five billion dollars
[23:33] Speaker 3: in funding its second round of financing this year, the
[23:37] Speaker 3: fast growing software firm, which competes with the likes of
[23:39] Speaker 3: Snowflake and Alphabet, also said that it crossed seven billion
[23:43] Speaker 3: dollars in revenue run rate, confirmation of a story I
[23:46] Speaker 3: think we've reported quite recently. We started talking today and
[23:50] Speaker 3: started the show on anthropics efforts to buy Dcar for
[23:53] Speaker 3: about six billion dollars according to sources, in what would
[23:56] Speaker 3: be its biggest acquisition yet. It comes as some of
[23:59] Speaker 3: the biggest A companies are spending to build and run
[24:02] Speaker 3: their models, and it raises questions about where the values
[24:05] Speaker 3: moving in AI. Right now, I've got an amazing person
[24:08] Speaker 3: to talk to you about that. Kak Will is general
[24:10] Speaker 3: partner at IVP, where she focuses on high growth companies
[24:13] Speaker 3: and the AI infrastructure layer. And it's great to have
[24:16] Speaker 3: you here on Bloomberg Tech.
[24:17] Speaker 6: It's great, too bad, Thank you.
[24:19] Speaker 3: I want to start with general question about buying and
[24:22] Speaker 3: building organically. You know, all layers of the five layer
[24:25] Speaker 3: cake stack. All of that's happening at once. Some people
[24:28] Speaker 3: are building, some people are out in the market. What
[24:30] Speaker 3: is it that you see most in your in your portfolio?
[24:34] Speaker 6: Yeah, good question.
[24:34] Speaker 12: So we're a venture capitalist, right, we're trying to invest
[24:37] Speaker 12: in all these companies we're you know, some of them
[24:39] Speaker 12: were going to sell to bigger companies.
[24:40] Speaker 6: But the ones that we love are the ones that.
[24:42] Speaker 12: Go all the way, right, the ones that go series
[24:45] Speaker 12: B to S and P as we like to say
[24:46] Speaker 12: at IVP.
[24:47] Speaker 6: So we're investing in the.
[24:48] Speaker 2: Likes series F to S and P, Series B, S
[24:51] Speaker 2: and P. That's that's the internal month.
[24:53] Speaker 6: I mean we joke.
[24:54] Speaker 2: Okay, sorry, I didn't meant to interrupt you. I just
[24:56] Speaker 2: like that.
[24:56] Speaker 11: No.
[24:57] Speaker 12: Yeah, so we're investing in the likes of the data
[24:59] Speaker 12: bre and to kartes, you know, hopefully those ten years
[25:03] Speaker 12: ago versus today. But yeah, I mean, I think it's
[25:05] Speaker 12: no surprise that in videos acquiring something like excuse me,
[25:09] Speaker 12: I said in video, IM ananthropic anthropic and acquiring a
[25:13] Speaker 12: Descarte because what you have in Decarte is you have
[25:15] Speaker 12: I mean, it's a world model. But I think that
[25:17] Speaker 12: the reason that they're acquiring Deacarte is because of this
[25:20] Speaker 12: inference optimization, and that is like one of these layers
[25:23] Speaker 12: of the stack that's becoming a super hot commodity.
[25:26] Speaker 3: Bof where to make the chips run better. In simple terms,
[25:28] Speaker 3: your infrastructure investor, your background is in software infra. Where
[25:33] Speaker 3: do you focus right now? What is your sort of
[25:35] Speaker 3: area of expertise in the lane that you're most focused on?
[25:38] Speaker 12: Yeah, so very much software infrastructure and cybersecurity.
[25:42] Speaker 6: Different reasons for the two.
[25:44] Speaker 12: Uh, you know, I learned this less lesson very viscerally
[25:47] Speaker 12: as a kid. Like my first job out of college
[25:49] Speaker 12: was working at Oracle selling databases, and in a single
[25:52] Speaker 12: day I was a sales rep. I visited Chevron into
[25:56] Speaker 12: It and Marmot, like the tech clothing brand, and I
[26:00] Speaker 12: was astounded at the rate, like sort of the breadth
[26:04] Speaker 12: of industries, but then the breadth of use cases too,
[26:07] Speaker 12: And that was the moment that it was like, you
[26:08] Speaker 12: know what, these infrastructure companies have vast markets and tams
[26:12] Speaker 12: and if you build something hard that no one else
[26:15] Speaker 12: wants to build, I mean, these can be long term
[26:18] Speaker 12: compounding businesses.
[26:19] Speaker 3: What are some of the sort of I don't know,
[26:22] Speaker 3: I guess you'd say you don't have a favorite child,
[26:24] Speaker 3: but in the portfolio, like which companies, yeah, are your
[26:28] Speaker 3: favorites right now that you see having the most potential.
[26:31] Speaker 12: Okay, well, let's sign us to public markets. And if
[26:34] Speaker 12: you look at the top ten trading names, you have
[26:36] Speaker 12: to get all the way to number eight or nine
[26:38] Speaker 12: before you get to a sort of a conventional software
[26:40] Speaker 12: sapas company, right, and one through eight are Pollunteer, crowd Strike,
[26:45] Speaker 12: Palo Alto, Snowflake, cloud Flare, et cetera. There's a reason
[26:50] Speaker 12: that these businesses trade so well, and it's because revenue
[26:54] Speaker 12: compounds with usage, right, and like implicit in that is
[26:58] Speaker 12: it is uncorrespond in two seats.
[27:03] Speaker 6: I've totally forgot the question.
[27:05] Speaker 7: Correct.
[27:05] Speaker 2: The question is which are your favorite?
[27:06] Speaker 12: Oh so my favorite my favorites are the ones that
[27:09] Speaker 12: do that, you know. So, my favorites are ClickHouse, My
[27:11] Speaker 12: favorites are Perplexity, a usage compounding business. My favorites are
[27:14] Speaker 12: Cribble laying, chain Base ten. All of these are technical
[27:20] Speaker 12: founders building technical products, selling into a technical audience. And
[27:24] Speaker 12: once you have them, the NAT dollar retention or like
[27:26] Speaker 12: the accumulating usage from those accounts is well over one
[27:30] Speaker 12: hundred one hundred and fifty percent.
[27:32] Speaker 2: Really interested in Perplexity.
[27:35] Speaker 3: You know, we've they seem to from a much broader
[27:38] Speaker 3: range of things now in their origin. One way that
[27:41] Speaker 3: you could explain to us is when when you make
[27:43] Speaker 3: the sponsorship or pitch to the rest of the partners
[27:45] Speaker 3: and why Perplexity, what is the future that you'd outline
[27:48] Speaker 3: to the team, why you'd backed them Perplexity.
[27:50] Speaker 6: It was easy. It was two things.
[27:52] Speaker 12: So September twenty twenty three, it's about two million an
[27:55] Speaker 12: arr as compared to the whatever Arvin has had publicly
[27:59] Speaker 12: five hundred million, so this is three years ago. It
[28:02] Speaker 12: came down to two things. Number one, the CEO, I mean,
[28:06] Speaker 12: Arvin is in the class of his own, a technical wizard,
[28:10] Speaker 12: extremely ambitious. And number two was as simple as they
[28:15] Speaker 12: ship product quickly. Yes, right, and that is a compounding
[28:18] Speaker 12: advantage in this environment. So I think the mistake with
[28:21] Speaker 12: Perplexity is thinking of it only as an answer engine,
[28:24] Speaker 12: just as the mistake with thinking about Anthropic as just
[28:26] Speaker 12: a model company. That short changes both of those companies dramatically.
[28:30] Speaker 6: Right.
[28:30] Speaker 12: Perplexity was the first to be an answer engine. It
[28:33] Speaker 12: was also the first to launch a browser comment. It
[28:35] Speaker 12: was the first to launch deep research, It was the
[28:38] Speaker 12: first to launch a computer agent with computer and so
[28:44] Speaker 12: that is a usage compounding story, right, because you put
[28:47] Speaker 12: all those things together and as revenue scales with.
[28:50] Speaker 3: Usage, we could sort of stop for any period of
[28:55] Speaker 3: time and say, there's a lot happening in AI.
[28:58] Speaker 2: Are we in an AI bubble? Et cetera, et cetera.
[29:01] Speaker 3: I would ask you what is the trend within AI
[29:03] Speaker 3: that you have identified? What do you think is happening
[29:06] Speaker 3: within all of the companies that you're backing that is
[29:09] Speaker 3: less talked about at the moment.
[29:11] Speaker 12: That they have unsatiable interest in demand?
[29:14] Speaker 2: But that's not changed, right.
[29:15] Speaker 3: Demand is still running far ahead of the industry's ability
[29:17] Speaker 3: to supply right exactly.
[29:19] Speaker 12: I think someone said this morning that they could sell
[29:21] Speaker 12: their whole twenty twenty seven book today if you know,
[29:25] Speaker 12: give them the chance.
[29:25] Speaker 6: Now, you wouldn't want to do that.
[29:27] Speaker 12: But I think the thing that people are missing is
[29:30] Speaker 12: we can simultaneously be in a golden age of technology,
[29:33] Speaker 12: which I think unequivocally I mean, I'm like diehard optimist,
[29:36] Speaker 12: but I think unequivocally we are, and we could have
[29:39] Speaker 12: a capital markets adjustment. I won't say correction, because that's
[29:43] Speaker 12: probably a technical term. But that wouldn't negate the fact
[29:47] Speaker 12: that it wouldn't say that the trend is fake. It
[29:49] Speaker 12: would just say that we have like a timing mismatch.
[29:52] Speaker 12: We probably have timing miss matches all the time. I
[29:55] Speaker 12: just think we're in the I mean, they're going to
[29:57] Speaker 12: be different phases of this sort of a proliferation, and
[30:01] Speaker 12: we're still in sort of the individual productivity phase.
[30:03] Speaker 6: So if you're a corporate CEO.
[30:04] Speaker 12: You might be underwhelmed, right because you're like, where is
[30:07] Speaker 12: it showing up in the numbers?
[30:09] Speaker 6: But if you look back, I mean, think of I
[30:11] Speaker 6: think it.
[30:11] Speaker 12: Was nineteen eighty seven, the MIT Nobel economist.
[30:15] Speaker 6: Richard Solo, he said he said the.
[30:19] Speaker 12: Computer age is showing up everywhere except in the productivity stets. Yes,
[30:24] Speaker 12: but that was in nineteen eighty seven, right, and like
[30:26] Speaker 12: look today. I mean he's also famous for the Solo paradox,
[30:30] Speaker 12: which said, like, which is really obvious today, but that
[30:33] Speaker 12: general purpose technology will show up in long term growth numbers,
[30:37] Speaker 12: which like unequivocally it has. And so I think right
[30:41] Speaker 12: right now we're probably in this like expectation mismatch, where
[30:44] Speaker 12: you have where you have it showing up in productivity,
[30:46] Speaker 12: but it's a little bit more individual and we needed
[30:49] Speaker 12: to flow through to workflows and then to just like
[30:51] Speaker 12: absolutely transforming companies.
[30:53] Speaker 3: I'd love to end the conversation by learning a little
[30:55] Speaker 3: bit about IVP. You know what the partnership's guiding principles are,
[30:58] Speaker 3: What you're trying to do differently?
[31:01] Speaker 12: Yeah, I mean, honestly, the thing we're trying to do
[31:03] Speaker 12: differently is that we have done the same thing for forty.
[31:05] Speaker 6: Six years, and that's just change.
[31:06] Speaker 12: If we can invest in ten companies of consequence a
[31:09] Speaker 12: year that have the potential to be generational companies, only ten, right,
[31:13] Speaker 12: and we don't want decision to fatigue. We invest in
[31:15] Speaker 12: ten companies a year, you know, that's our issues.
[31:18] Speaker 3: Is it harder to resist the urge tend to invest?
[31:21] Speaker 6: Absolutely? But venture is a picking business.
[31:24] Speaker 12: Venture is not a index coverage business. And hence we
[31:29] Speaker 12: stick to our We do one thing, one thing well,
[31:31] Speaker 12: and that's invest in ten companies a year.
[31:35] Speaker 3: It's okay based on reporting. Therefore, if some funds have
[31:39] Speaker 3: fifty percent write down, you just say fifty percent of
[31:42] Speaker 3: these didn't work out. I'm asking for an insight to
[31:46] Speaker 3: the reality of how that works at the end.
[31:48] Speaker 12: Oh yeah, I mean in venture you have to adjust
[31:54] Speaker 12: quickly that you can't flog yourself for the losers, because
[31:58] Speaker 12: the losers are in service of the winners. Right across
[32:01] Speaker 12: my portfolio, probably one or two companies will make up
[32:06] Speaker 12: the vast nugget of returns versus the long tail.
[32:11] Speaker 3: I guess my question was that's still true in the
[32:13] Speaker 3: AI age.
[32:14] Speaker 12: Oh, that is more true today than it was yesterday,
[32:17] Speaker 12: and it's five times more true than it was five
[32:19] Speaker 12: years ago.
[32:20] Speaker 6: Yeah.
[32:21] Speaker 12: Yeah, I mean look at this morning, right, Lovable raising
[32:24] Speaker 12: at thirteen billion and Cognition.
[32:27] Speaker 6: Raising at forty billion.
[32:28] Speaker 12: I think there are these just like crowned winners, and
[32:32] Speaker 12: those are going to be the ones that return all
[32:34] Speaker 12: the returns in menticapit and I think what.
[32:35] Speaker 3: You said earlier is look at the companies that can
[32:37] Speaker 3: ship product. Kaquil MVP. It's been great to have you here.
[32:41] Speaker 3: Thank you, Vanessa, thank you very much. Indeed, now coming
[32:43] Speaker 3: up a threbra shares slide off, the company projected slower
[32:46] Speaker 3: growth then some investors had participated. Some balls out there
[32:50] Speaker 3: wanted more. We'll speak to CEO Andrew Feldman. He joins
[32:54] Speaker 3: us next. This has been big tech Okay for just
[33:02] Speaker 3: the second time since its ipo. Cerebris reported quarterly earnings
[33:06] Speaker 3: yesterday after the closing bell, offering some investors' growth that
[33:10] Speaker 3: was slower than anticipated. Shares down thirteen point four percent.
[33:15] Speaker 3: I think that's on track for the biggest drop since
[33:17] Speaker 3: the last week of June. Cerebra CEO Andrew Feldman joins us. Now,
[33:22] Speaker 3: the first thing that I learned was there is a
[33:26] Speaker 3: business renting out compute capacity that's still there.
[33:30] Speaker 2: Could you give it a backstory on that? Andrew, Oh, of.
[33:34] Speaker 11: Course, there's a huge business of renting out compute capacity.
[33:38] Speaker 11: I think it is. It's what the neo clouds do,
[33:42] Speaker 11: it's what the hyperscalers do.
[33:45] Speaker 7: It is a it's a very profitable business.
[33:50] Speaker 3: And is he outpacing the growth from the business of
[33:53] Speaker 3: selling the hardware direct to others.
[33:56] Speaker 11: Well, we have two parts of our business. We have
[34:00] Speaker 11: a first party cloud that means we design, manufacture and
[34:06] Speaker 11: deploy equipment in our own data centers, and we rented
[34:09] Speaker 11: out to our customers.
[34:10] Speaker 7: That part of the business was up two.
[34:13] Speaker 11: Hundred and eighty one percent year over year, huge growth,
[34:17] Speaker 11: huge growth. Now on that part of the business, we
[34:19] Speaker 11: control the data centers. There was another part of our
[34:23] Speaker 11: business where we sell our AI hardware to customers who
[34:30] Speaker 11: want to do the exact same thing, who want to
[34:32] Speaker 11: deploy the equipment and either use it for themselves or
[34:39] Speaker 11: break it up into little pieces and sell the compute
[34:42] Speaker 11: to various customers. Right those customers, we call that our
[34:47] Speaker 11: hardware business. We're selling hardware for their premise and in
[34:51] Speaker 11: Q two some of our customers didn't have data centers available,
[34:55] Speaker 11: so we were able to deliver less to them. And
[34:59] Speaker 11: so right now, data centers are a challenge for everybody.
[35:02] Speaker 7: We had some we were able to deploy.
[35:05] Speaker 11: Them for our own use and for the use of
[35:08] Speaker 11: our cloud customers, our first party cloud customers, and so
[35:12] Speaker 11: that business is absolutely ripping.
[35:16] Speaker 3: We're live on Bloomberg Television and Radio with three BRIS
[35:18] Speaker 3: CEO Andrew Feldman.
[35:20] Speaker 2: And the reason I asked for the.
[35:21] Speaker 3: Backstory on that is that that cloud business was borne
[35:23] Speaker 3: out of needing to build data center capacity to test
[35:26] Speaker 3: the technology worked it seems to be working.
[35:30] Speaker 11: It was, you know, we wanted to be able to
[35:33] Speaker 11: provide demonstrations to customers without having to move hard work
[35:37] Speaker 11: to their facilities. We didn't have a hyperscale customer at
[35:44] Speaker 11: that time, and so there was a need to find
[35:47] Speaker 11: a way to deliver compute services to people who didn't
[35:53] Speaker 11: have a data center, didn't have a way to otherwise
[35:56] Speaker 11: enjoy our blisteringly fast AI hardware, and so we set
[36:01] Speaker 11: it up at first for ourselves and then for demonstrations,
[36:05] Speaker 11: and there was so much demand on it over time
[36:08] Speaker 11: that it's become a big business for us. And last
[36:11] Speaker 11: quarter Q two it was an extraordinary.
[36:14] Speaker 7: Business for us.
[36:15] Speaker 11: And if we expected to continue to grow at tremendous rate.
[36:19] Speaker 11: Now even though now we have hyperscale customers like a WS,
[36:23] Speaker 11: now that we have frontier labs like open Ai, that
[36:27] Speaker 11: business continues.
[36:29] Speaker 7: To grow at exceptional rates.
[36:31] Speaker 3: Okay, the hardware business, there is some confusion out there.
[36:35] Speaker 3: You report on a core basis, which showed that the
[36:38] Speaker 3: hardware division was up right, on a gap or non
[36:42] Speaker 3: core basis, it would be otherwise.
[36:45] Speaker 2: Just explain the story for the hardware business.
[36:48] Speaker 7: The story is.
[36:51] Speaker 11: One of our large hardware customers had some warrants and
[36:55] Speaker 11: those warrants are subtracted from revenue.
[36:58] Speaker 7: The value of those warrants are.
[36:59] Speaker 11: Subtracted from revenue in Gaba County, and so.
[37:03] Speaker 7: It made the numbers look a little smaller than they were.
[37:07] Speaker 3: Actually, demand is great, it's there, there is a background.
[37:15] Speaker 11: We know about it being and we've got twenty four
[37:18] Speaker 11: billion in back lass.
[37:20] Speaker 3: So here is the data that people are really searching
[37:23] Speaker 3: for with Cerebras, which is how can you get the
[37:25] Speaker 3: capacity in place faster, both on your own data centers
[37:29] Speaker 3: or to get the hardware into the hands of customers
[37:32] Speaker 3: for theirs.
[37:34] Speaker 7: And that's a fair question.
[37:35] Speaker 11: What we said to on our coyers today is that
[37:38] Speaker 11: we had signed up or had already live six hundred
[37:42] Speaker 11: megawatts of capacity, and that we would have that we
[37:48] Speaker 11: had a pipeline of gigawatts of additional capacity. And what
[37:53] Speaker 11: we said was that we expect our revenue to more
[37:56] Speaker 11: than triple next year. And so I think we have
[38:00] Speaker 11: been executing at a tremendous rate.
[38:04] Speaker 7: You know, adding six hundred.
[38:05] Speaker 11: Megawatts in several months, six or seven months is no
[38:09] Speaker 11: small feat, and we will add add vastly more in
[38:14] Speaker 11: the months to come, and so.
[38:16] Speaker 7: I think we are ahead of plan.
[38:21] Speaker 11: We beat guidance and we beat consensus on the things
[38:25] Speaker 11: we on our core business, on core revenue, core gross margins,
[38:31] Speaker 11: and beat by a lot on core operating margins, and
[38:34] Speaker 11: then we guided a big raise so for twenty twenty
[38:39] Speaker 11: seven and twenty eight, So.
[38:41] Speaker 7: We feel really good about the business and.
[38:47] Speaker 11: Are proud of of of last quarters results.
[38:52] Speaker 3: Andrew, what are the biggest factors on the supply chain
[38:54] Speaker 3: side or on the operations side that are impacting your
[38:58] Speaker 3: ability to deploy that capacity?
[39:01] Speaker 11: So there are several things you need to do on
[39:03] Speaker 11: the supply chain right First, we just talked about data centers.
[39:08] Speaker 11: You need a place to put your equipment so you
[39:11] Speaker 11: can build the cloud so that you can deliver to
[39:13] Speaker 11: your customers. Walking back upstream, you need manufacturing capacity, and
[39:22] Speaker 11: we've expanded our capacity by partnering with Electronics and sand
[39:28] Speaker 11: Mina and rocket Ems. We will this year increase manufacturing
[39:34] Speaker 11: capacity by more than ten x. Walking further back upstream,
[39:39] Speaker 11: you have to work with your fab partner.
[39:41] Speaker 7: Our partner is TSMC, and.
[39:44] Speaker 11: They have assured us supply that will support our growth.
[39:48] Speaker 11: And so we've systematically worked each of these components in
[39:53] Speaker 11: a field we call capacity to have more chips, more
[39:58] Speaker 11: manufacturing capacity, more data centers to put these chips, to
[40:02] Speaker 11: build the to build the cloud, offering.
[40:05] Speaker 3: The socks down fourteen percent right now, as you and
[40:08] Speaker 3: I are talking. You know, since the IPO it's performed.
[40:12] Speaker 3: I think this STOT was up forty two percent after
[40:14] Speaker 3: from tesday's close. Is lock up expier factor here? You
[40:18] Speaker 3: know we talked earlier in the program about SpaceX's lock
[40:21] Speaker 3: up and the mechanics of that. Would you say that
[40:23] Speaker 3: you see an impact of that in your case?
[40:26] Speaker 7: We do.
[40:26] Speaker 11: They are lockups that are that are expiring, and that
[40:29] Speaker 11: there are people who invest against those. Those aren't. Those
[40:36] Speaker 11: aren't the investors that we're focused on. We're focused on
[40:40] Speaker 11: long term investors. We're focused on delivering returns through investing
[40:46] Speaker 11: in our people, in our technology that will that will
[40:50] Speaker 11: play out over years and when you when you guide
[40:54] Speaker 11: to more than a triple next year and say that
[40:58] Speaker 11: twenty eight is going to be going to be giant
[41:00] Speaker 11: and twenty nine as well. We're trying to speak to
[41:04] Speaker 11: people who want to be with us on this ride
[41:07] Speaker 11: and who want to be in the AI, in the
[41:11] Speaker 11: AI business and want alternatives to say in video, and
[41:20] Speaker 11: that's who we're talking to.
[41:23] Speaker 2: Andrew.
[41:23] Speaker 3: I appreciate it's only been a few weeks since this
[41:25] Speaker 3: was announcing you were on the show, But how's it
[41:27] Speaker 3: going with AMD engineering teams already working together.
[41:32] Speaker 2: Have you moved quickly on that.
[41:35] Speaker 11: I enjoyed that conversation when you had me on your
[41:37] Speaker 11: show a few weeks ago.
[41:38] Speaker 7: I appreciate that. And it is you know, Lisa Sue
[41:43] Speaker 7: and Mark paper Master. They acquired my last company.
[41:46] Speaker 11: And I had an opportunity to work with them for
[41:47] Speaker 11: several years and I enjoyed it a great deal. We
[41:51] Speaker 11: have a good relationship there and we have disaggregated solutions
[41:55] Speaker 11: running in our labs. We are we have close engineering relationships.
[42:00] Speaker 11: These our talented engineers on both sides working extremely hard
[42:04] Speaker 11: and we will have deployments in the fourth quarter.
[42:08] Speaker 3: And very quickly. There was customer demand after the announcement.
[42:12] Speaker 7: There is enormous custom demand.
[42:14] Speaker 11: Disaggregated solution provides the best of both worlds.
[42:17] Speaker 7: It allows GPUs to go faster. GPUs are good at
[42:22] Speaker 7: throughput and not so good at speed.
[42:24] Speaker 11: They allow GPUs to go faster and cerebris which is
[42:28] Speaker 11: really good at speed, and we can improve and throughput.
[42:31] Speaker 11: And so the disaggregated solution gives a faster, high throughput
[42:34] Speaker 11: solution and that gives you both the economics which is
[42:38] Speaker 11: throughput and the customer experience, which is speed.
[42:43] Speaker 3: Threebras CEO Andrew Feldman back on Bloomberg Tech, thank you
[42:46] Speaker 3: very much.
[42:47] Speaker 2: Indeed having me a.
[42:48] Speaker 3: Meta says it has shut down more than seven hundred
[42:51] Speaker 3: and fifty thousand Instagram and Facebook accounts in Australia believed
[42:55] Speaker 3: to belong to users under sixteen, to comply with the
[42:58] Speaker 3: country's landmark social media ban. The band's being watched closely
[43:02] Speaker 3: to see if it curbs social media use among young
[43:04] Speaker 3: people that will. Report last month by the Australia's Online
[43:07] Speaker 3: Safety Reguator said it led to only a marginal reduction
[43:11] Speaker 3: after taking effect in December. That brings us to another
[43:15] Speaker 3: big number, one point four trillion dollars. That's the potential
[43:19] Speaker 3: financial penalty. Meta says it could face in a landmark
[43:23] Speaker 3: social media addiction trial kicking off in California next week.
[43:28] Speaker 3: A coalition of twenty nine state attorneys general alleged Meta
[43:32] Speaker 3: engaged in deceptive business practices affecting millions of young users
[43:37] Speaker 3: and that led to a mental health crisis. Let's get
[43:40] Speaker 3: the details with Bloomberg's legal reporter Madeline Meckelberg. You know
[43:44] Speaker 3: it's a big number, the potential penalty, but this is
[43:48] Speaker 3: us teeing up for something to come this week.
[43:52] Speaker 2: What do we need to know? That's right.
[43:55] Speaker 10: It's a massive number, especially when you consider what Meta's
[43:58] Speaker 10: market value is, which is one point five trillion dollars.
[44:02] Speaker 10: So if this figure actually came to pass, it would
[44:05] Speaker 10: be staggering and record setting in the realm of civil detigation.
[44:09] Speaker 10: I think it would be the highest penalty ever levied.
[44:12] Speaker 7: But you're right.
[44:12] Speaker 10: The trial started this week with jury selection, but next
[44:16] Speaker 10: week we're actually going to get into the substantive arguments
[44:19] Speaker 10: when we're going to hear opening arguments from both sides
[44:22] Speaker 10: about this case, and then we're going to kick off
[44:24] Speaker 10: in about a month for about a month of testimony.
[44:27] Speaker 10: We're expecting to hear from Meta CEO Mark Zuckerberg and
[44:30] Speaker 10: other executives at the company, as well as folks from
[44:33] Speaker 10: these state attorneys general's offices who are going to talk
[44:36] Speaker 10: about what we can expect or what they're saying is
[44:39] Speaker 10: the impact that these policies that Meta have had on
[44:42] Speaker 10: individuals in their state.
[44:44] Speaker 3: Madlin, What is literally the legal issue that will be
[44:48] Speaker 3: debated here.
[44:50] Speaker 10: This is an interesting case because it's different than one
[44:53] Speaker 10: that we've talked about before. Loyal viewers will remember the
[44:56] Speaker 10: case that we heard about in Los Angeles earlier this year.
[45:00] Speaker 10: This is about a bigger question than just harm to
[45:03] Speaker 10: an individual user. This is about harm to a broader population.
[45:08] Speaker 10: As you said, they're out alleging a mental health crisis,
[45:11] Speaker 10: and they've said that these algorithms and these tools that
[45:14] Speaker 10: are being employed by the companies are contributing to widespread issues.
[45:19] Speaker 10: It's under this It's compared often to the litigation against
[45:23] Speaker 10: Big Tobacco, because it's about whether or not this company
[45:27] Speaker 10: is considered a public nuisance and is causing harm to
[45:31] Speaker 10: individuals in a broad population.
[45:34] Speaker 3: Very quickly, Maden, Just this morning we've heard from Meta
[45:37] Speaker 3: what are they saying going into the trial.
[45:42] Speaker 10: Metas kind of maintained the same line across all of
[45:44] Speaker 10: these cases, you know, As I said, this is one
[45:46] Speaker 10: of several cases, but of course a very significant one,
[45:49] Speaker 10: and they've said that they have safety features that they
[45:52] Speaker 10: believe helped make these experience for young users safe and
[45:56] Speaker 10: that their work has been sufficient.
[45:58] Speaker 3: Blienbog's, Madam Meckelburg, thank you for much. From breaking news,
[46:01] Speaker 3: open AI is naming a chief new chief revenue officer
[46:04] Speaker 3: for the second time in less than a year. This
[46:07] Speaker 3: comes as an effort to bolster sales growth ahead of
[46:10] Speaker 3: an highly anticipated Wall Street debut. AI startups high Dally Rajitch,
[46:14] Speaker 3: president and Chief operating officer of the Alphabet owned cybersecurity
[46:17] Speaker 3: firm Whiz, as its new revenue chief, succeeding Denise Dresser,
[46:21] Speaker 3: who leaves open Ai to pursue other opportunities following a
[46:25] Speaker 3: transition period. That does it for this edition of Bloomberg Tech.
[46:29] Speaker 3: Recount the show on the podcast New where to find
[46:31] Speaker 3: it on the Terminal, as well as online on Apple, Spotify,
[46:34] Speaker 3: and iHeart This is Bloomberg

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.
